Title : NSF 93-69 - Summary of NSF Workshop on Research Opportunities in Manufacturing in the Process Industries Type : Report NSF Org: ENG / CTS Date : May 7, 1993 File : nsf9369 ****************************************************************************** This File has been updated 10/31/96 to reflect the proper address of the: National Science Foundation 4201 Wilson Boulevard Arlington, VA 22230 For more information call: (703)306-1234 ****************************************************************************** OPPORTUNITIES IN MANUFACTURING RESEARCH IN THE PROCESS INDUSTRIES Summary of a National Science Foundation Workshop December 2-4, 1992 Edited by Henry A. McGee, Jr. Division Director, Chemical and Transport Systems Division and Maria K. Burka Program Director, Chemical Reaction Processes Program Chemical and Tansport Systems Division National Science Foundation Washington, DC 20550 TABLE OF CONTENTS PREFACE 2 INTRODUCTION 3 REPORT OF THE MICROELECTRONICS BREAK-OUT GROUP 5 REPORT OF THE AUTOMOTIVE BREAK-OUT GROUP 18 REPORT OF THE CHEMICAL INDUSTRY BREAK-OUT GROUP 29 CONCLUDING REMARKS 39 AGENDA 43 LIST OF ATTENDEES 45 ABSTRACTS OF PRESENTATIONS BY PARTICIPANTS 48 PREFACE An NSF workshop on Manufacturing in the Process Industries was held on December 2, 3, and 4, 1992. The purpose of the workshop was to bring together industrial practitioners, academic researchers, and government administrators to reflect on and discuss the future role of the Engineering Directorate (ENG), with emphasis on the process industries, in a possible government-wide initiative in research on manufacturing. The meeting was held at the National Science Foundation, 1800 G Street, N.W., Washington, DC 20550. Each participant was asked to give a short presentation on experiences and thoughts on the potential impact of advances in fundamental research on manufacturing. Some of the participants supplied an abstract of their presentations. The workshop agenda, list of participants and these abstracts are appended. Dr. Maria Burka gave a brief overview of the rationale for the workshop. Dr. Joseph Bordogna, Assistant Director for the Engineering Directorate (AD/ENG), then gave a summary of the present state of the proposed manufacturing initiative at the Foundation. Dr. Henry McGee then outlined the scope of the research done in the CTS Division and presented some of his own ideas of how manufacturing could be integrated into ongoing Divisional activities. Dr. Marshall Lih discussed the work of the Engineering Research Centers, with specific emphasis on the manufacturing research done in selected ERCs. Mr. John Pfeiffer presented an overview of the Federal Coordinating Council for Science, Engineering, and Technology (FCCSET) process from the viewpoint of the Office of Management and Budget (OMB). Dr. Stanley Abramowitz discussed the various activities in the area of manufacturing by the National Institute of Standards and Technology (NIST). These introductory presentations were followed by presentations on specific manufacturing issues in the metals, automotive, microelectronics, and chemical industries (abstracts are appended). These talks constituted the primary input into the subsequent deliberations of the panel. These industries are representative (but not inclusive) of the many industries that could benefit from research advances in manufacturing technologies. After these presentations, break-out groups were formed according to the various industrial sectors that were represented. The report of the various break-out groups is the main substance of this workshop report. It is important to note that this document is a record of the thoughts of this particular group of people and is not a position paper reflecting NSF or ENG policy. In each subsection locally accepted "jargon" is used. For example, a chemical reactor is called a "unit operation" in the chemical industry while it is called a "tool" in the microelectronics industry. INTRODUCTION A Federal Coordinating Council for Science, Engineering, and Technology (FCCSET) committee has been meeting over the past few months to explore the role that government agencies can play in advancing manufacturing in the U.S. In charging this FCCSET group, the then Science Advisor Allan Bromley stated, among other remarks: "A strong manufacturing base is vital to the economic and military strength of our nation. This program will leverage the world class technological capabilities of the United States Government to address the manufacturing needs of a broad sweep of industrial sectors." "The problems facing our manufacturing industries are complex: some are technology-related, while others, like changing management requirements, are clearly the responsibility of the private sector." While the manufacturing FCCSET group held its meetings, various subgroups within NSF also began to explore the role different portions of the Foundation should be playing in this enterprise. The Engineering Directorate (ENG) set up a Strategic Planning Subcommittee to explore what contributions ENG could make and how best to achieve the goals of an initiative in this area. The first step was to determine what constitutes "manufacturing". A definition that was developed at that time was: "Manufacturing: the creation and integration of informational and physical processes to create economic wealth through the production of artifacts. This includes the design of products, manufacturing processes and facilities; the physical processing of materials into products; and processes associated with the product life cycle." The underlying problem was determined to be that "the competitive nature of contemporary human society dictates that only those industrial enterprises that can rapidly bring to market high quality and innovative products at the lowest cost will command a major market share and be profitable." The group then began to explore what strategic opportunities exist for NSF to impact U.S. manufacturing. Specifically, how "NSF, in close cooperation with the academic community, can provide a leadership role in fashioning the education and research activity in manufacturing to enable the nation's industrial sector to lead the world in responding best to market forces." Research opportunities related to manufacturing exist throughout NSF and throughout ENG. While programs exist which fund manufacturing in small and disparate ways, optimal impact is only possible if a concerted and coordinated effort is initiated, which embraces new paradigms and combines them with the relevant portions of existing programs. It was the purpose of this workshop to provide input into this overall process. REPORT OF THE MICROELECTRONICS INDUSTRY BREAK-OUT GROUP INTRODUCTION The goals of this section are to generate ideas which will advance and improve microelectronics manufacturing, by way of reduced time to market for new products and processing techniques as well as added value for existing technology through robust/agile scheduling algorithms, improved product quality and reduced cost. The text on the following pages captures a set of inputs on science/engineering issues that are relevant in this context. They are intended as support of proposal solicitation in this area. A list of topics on various aspects of integrated circuit (IC) manufacture is provided, followed by supplementary remarks highlighting the items deemed especially important and/or accessible to joint efforts by academe and industry. The bulk of this report addresses device, process or tool related concerns having to do with film deposition, removal and cleaning operations. By way of disclaimer, it is acknowledged that some items relating to optoelectronics, display technology, IC packaging and lithography have been omitted. It was felt that IC packaging is generally more mature than the technologies listed in the following subsections and lithography demands expertise in optics, radiation and beam physics not represented by the review panel. Similarly, the area of bulk silicon prod! uction as wafer material has been omitted, in keeping with current interpretations of the 'microelectronics scope', though it is hoped that this subject will be adequately covered by other NSF initiatives. The panel recommends a proposal format that involves cooperative industry/academic research. In favoring joint proposals by academe and industry, NSF will encourage active involvement, i.e. investment of time and effort, from industrial staff in collaboration with university faculty, and strongly support effective technology transfer (in the broadest sense). The latter is a sine-qua-non for this Manufacturing Initiative. In fact, there was strong consideration of asking the industrial partner to furnish a cost-benefit analysis for the proposed work but there was considerable controversy on this issue and no consensus was reached by the group. It is also suggested that when drafting a program announcement for such a Manufacturing Initiative, historical knowledge gained from the progress of various SBIR programs and the NSF/DARPA cooperative effort on VLSI in the 1980's be considered as useful role models. OUTLINE (OVERVIEW) OF RESEARCH NEEDS IN MANUFACTURING -MICROELECTRONICS INDUSTRY - not prioritized General Research Topics 1. Fundamental Chemistry/Physics Knowledge Base -Chemical kinetics -Reaction pathways 2. Processes and Process Integration -CVD and MOCVD -Plasma etching, deposition -RTP -PVD -Cleaning -Sources/delivery/disposal -Liquids: resist materials, liquid crystal materials, CVD source precursors 3. Equipment -Design for process embodiment and reliability - defects -Wall reactions -Ultra clean design -Clusters and tool integration -Multifunction tool modules/ ASWP -Novel tool concepts 4. Technology Modeling: heat, momentum, chemical species -Tool scale: reactor, source, delivery, exhaust -Microstructure-dependent effects, profile development -Boundary layer -Atomic scale: microstructure, morphology, properties -Tool systems (clusters) -Sensor-based control -Integration with design of experiments -Thermal and enhanced (ion, radiation) interactions 5. Sensors and Control -In-situ, real-time -Control models and algorithms -In-line and off-line 6. Ultra clean Processing -Chemistry and physics of reactive impurities -Particle nucleation, transport, removal -Defect identification and characterization Necessary Characteristics of Research Topics for Optimal Impact in Microelectronics Manufacturing 1. 3-D Understanding and control of Processes -Microfeatures, profiles -Pattern factors/effects -Tool scale uniformity -Aspect ratio dependent effects -Material selectivity 2. Modeling -Experimental model validation/support -Link to broad utilization/commercialization: benchmarks, CAD -Process/equipment design by modeling 3. Equipment -Smart tools -Process control and optimization -Tool standardization and simplification-Process and tool integration -Design for reliability and predictability 4. Defects -Mechanisms of generation -Identification -Control -Competing reactions 5. Environmental Aspects -Manufacturing floor: safe process and tools -Local and global environment/ waste disposal, minimization -Energy usage -Natural resource usage 6. Reliability DISCUSSION OF RESEARCH TOPICS/NEEDS MICROELECTRONICS INDUSTRY The understanding and control of processes and chemistry are very critical areas for VLSI and ULSI-based silicon integrated circuits. The following are examples (by no means a comprehensive list) of specific systems of interest and specific challenges they present. Nearly all of the processes require understanding from the point of view of how the local feature topography affects processing requirements (aspect ratio dependence). The focus should be to induce manufacturing improvements. Fundamental Chemistry/Physics Knowledge Base Measuring and calculating chemical kinetics, reaction pathways, reaction cross sections, etc. for key reactions on silicon wafer surfaces during device fabrication provide understanding and input to modeling. Reactions of importance can be obtained from the following section. A prioritized list of reactions is not in the scope of this write-up. Processes and Process Integration 1. CVD and MOCVD a. Deposition of amorphous and polycrystalline silicon on silicon substrates/devices: electrical and microstructural properties control b. Silicon dioxide deposition in high aspect ratio features: poor profile control in trenches and holes (voids are left), quality of films are often poor or non uniform (chemical stability, mechanical stress), new chemistries can be examined c. Selective deposition on microfeatures including metals on seed layers, barriers on seed layers, silicon on silicon: finite selectivity and cleanliness of surface cause excessive defects (growth where it is not wanted), morphology and properties not optimum with processing conditions 2. Plasma Etching and Deposition a. Aspect ratio dependence of etching and deposition: this spans most of the important processes in manufacturing, how to control it for each system. For etching the metric needed is control of profile and material selectivity. For deposition it is uniformity of film thickness and quality of film properties. b. New chemistries, plasma generation methods and processes for etching and deposition which offer improvements in film properties, control of profiles, ease of fabrication, etc. c. Charge-up/radiation-induced damage to electrical devices and how to avoid or to minimize them: Advanced plasma sources driving high rate processes over very thin device regions have a great chance of creating damage which cannot be annealed out or removed, particularly for the overetching step in an etching process. Control of the plasma physics and the process chemistry are crucial. Charge-free etching processes with energetic, but directed neutrals may be required and this is a new area of research for manufacturing. d. Identifying and understanding the origins of particles -control in this area is particularly important for plasma processes. 3. Rapid Thermal Processing (RTP) a. Heat sources necessary for rapid and uniform heating of 200 mm to 300 mm wafers. b. Measurement and control of temperature uniformity in heating cycles of short duration with rapid transients (order of 100 C/sec) c. Rapidly-switched gas injection for short, uniform exposure of wafer surface to reactants. 4. Physical Vapor Deposition (PVD) a. High-rate deposition in high aspect ratio features (collimated sputtering, other ways to direct momentum transport?) b. Sputtering target uniformity and transport to wafer, target life and target purity. c. Pattern-dependent deposition 5. Surface Cleaning and Preparation (Cleaning can constitute 50% of steps to make devices) a. Chemical vapor phase treatment of surfaces (especially dry processing) b. Plasma cleaning: chemical or physical (sputtering) c. Removal of particles left from processing 6. Sources, Delivery and Disposal of Chemicals a. Purification, transport, characterization of feedstock materials for deposition, etching, cleaning, etc. b. Feed point versus uniformity and % utilization in reactors c. Point of use generation of chemicals d. Use or conversion of exhaust flow, environmental concerns 7. Liquids used in Processing a. Purity of materials and ability to recycle used materials b. Feedstock chemicals for deposition, their handling during processing c. Liquid crystal materials processing Equipment 1. Design of reactors for specific processes 2. Clustering of processes in integrated tools 3. Tool simplification - multi-use reactor designs 4. Equipment- and process-related defect sources 5. Ultraclean process equipment Technology Modeling - heat, momentum, chemical species 1. Tool scale modeling of process reactor a. Source generation b. Transport c. Boundary layer 2. Microfeature scale modeling a. 3-D profile evolution b. Surface kinetics c. Reaction pathways 3. Atomic scale modeling 4. Thermal and enhanced (ion, radiation) interactions 5. Applications to design of experiments, sensitivity analysis 6. Applications to sensor definition Sensors and Control 1. In-situ, real-time sensors 2. Model-based control 3. In-line measurement and characterization 4. Off-line diagnostics and metrology 5. Defect detection and failure analysis Ultraclean Processing 1. Reactive impurities 2. Particle nucleation, transport, removal 3. Sparse defect characterization 4. Reagent purity DISCUSSION OF CRUCIAL CHARACTERISTICS OF RESEARCH TOPICS FOR MICROELECTRONICS MANUFACTURING IMPACT Many research topics already supported by NSF CTS can benefit competitive manufacturing in microelectronics. The previous section provided a detailed perspective on which specific issues are most important. This section attempts to abstract central themes which would add substantial value to CTS research programs aimed at manufacturing applications. It is important to note the distinction between affecting today's manufacturing and that of tomorrow. There certainly are areas where technical expertise can increase today's manufacturing effectiveness, i.e. on a short time scale. However, systematic benefits to manufacturing competitiveness will usually require investigation and demonstration exercises of new process and equipment approaches prior to their implementation in manufacturing. This means that enhancing manufacturing may often translate into adding value to the manufacturing approaches defined during development phases of a new technology generation. A time scale of several years for impact on the target is consistent with the notion of engineering or manufacturing research, as might be implied by its presence in the university environment and support by NSF. In addition, because of the nature of research itself, much of its impact on manufacturing is likely to take the form of enhanced manufacturability. This means understanding and controlling processes and equipment at a considerably more demanding level than needed for conventional development and demonstration purposes. Rather, manufacturability implies understanding and control sufficient to do such things as accurately assess process windows, predict statistical variations related to equipment design, and gauge the importance of defect generation mechanisms such as those associated with the presence of competing reaction pathways. 1. Three Dimensional (3-D) Understanding and Control of Processes The goal of microelectronics manufacturing is to fabricate complex 3-dimensional structures with precise thin film geometries. This represents a significant step beyond formation of high quality, blanket thin films. Success requires a more demanding level of understanding and control of the process chemistry and physics involved, as well as its embodiment in equipment and its consequences in the properties and stability of the thin film material structure which results as the manufacturing product. The most obvious length scale at which 3-D fabrication must be controlled is that of the microelectronic structures, namely in the submicron range. Pattern definition is achieved by lithographic processes, following which plasma-based material removal by etching is carried out. The 3-D profile of etched thin film structures plays a crucial role. Process chemistry and physics determine the relative degree of anisotropic (directional) vs. isotropic etching occurs. For transferring a minimum-feature-size lithographic pattern (e.g., FET gate definition) into thin film material, lateral geometry must be retained through essentially vertical etching (100% etch directionality). In other cases tapered profiles (wider at top than bottom) may be desired, e.g. to prevent void formation upon subsequent CVD filling of the structure with material, and a less directional etch is needed. Similar considerations arise in deposition, where the process must often lead to material deposition on sidewalls and bottom of microelectronic features effectively. For example, high conformality (step coverage) can produce sidewall insulator thicknesses essentially as large as the top coverage, with vertical sidewalls and spatially homogeneous properties of the thin film. In many cases a re-entrant profile (sidewall coverage higher at the top of the trench than at the bottom) is essential, e.g. to prevent formation of voids when complete filling of the trench with material is attempted. The specific chemistry and physics employed in the process determines the 3-D profile. Understanding and control of the process is thereby crucial, and furthermore for manufacturing it is essential to verify that the process window for achieving acceptable profiles from the chemistry and equipment exists. Mechanisms which determine 3-D profiles in etching and deposition are currently not well understood, so research in this area should be a high priority. Corresponding considerations apply on the length scale of the process equipment (tools), for uniformity of etching or deposition across large (8") wafers is essential in order to achieve sufficient yield and low cost. Uniformity issues involve tool design as well as its interaction with local chemistry and physics. Thermal distributions in process reactors, as well as flow patterns and concentrations of chemical species, determine wafer-level uniformity. Tool design becomes a major challenge, and its optimization is then inextricably linked with the specific process chemistry being employed. Other considerations arise at intermediate length scales. First, microfeature profile control can be heavily dependent on the aspect ratio (depth/width) of trench features. Aspect ratios must become larger with new generations of technology, thereby forcing more difficult challenges for 3-D profile control. In DRAM trench capacitors, aspect ratios of at least 40 are needed (e.g., 10 um depth x 0.25 um width). In addition, the dependence on chemical reactions - with corresponding volatile reaction products - and thermal/flow distributions leads to changes in 3-D profiles and reaction rates depending on the immediate environment over mm length scales; this is manifested as pattern-dependencies, such as deeper etching of trenches where trench density is lower cf. the behavior in areas of high trench density. At the tool scale, loading effects can play an analogous role: in batch reactors where reactant depletion influences the reactor conditions needed for all wafers to! be modified in the same way, changing the number of wafers in the batch leads to wafer-to-wafer nonuniformity. Microelectronic structures involve not only 3-D topography on a submicron scale, but also the presence of different materials, which respond differently to particular process chemistries. In many cases efficacy of profile fabrication in manufacturing depends on understanding and control of reaction selectivity, which may be exploited in order to terminate etching at an interface, deposit only in some regions of the wafer surface, etc. Again the 3-D character of the microfabrication process depends on understanding and control of chemical and physical phenomena. It is worth noting that significance for manufacturing is also attached to 3-D profile control on an atomic or nm scale. This comment applies particularly to materials which are not large single-crystals. In semiconductors, the morphology and microstructure of undoped or doped films is determined by process chemistry and physics which is not well understood, with applications to thin film transistors (e.g. for active matrix displays), solar cells, and polySi interconnections in VLSI. Microstructure and morphology arising from deposition or etching is also crucial in properties of metals (e.g., from W CVD) or insulators. Microelectronic structure fabrication necessitates 3-D process control on the above length scales. Successful manufacturing of such structures at very high density and integration level increases the demands on process knowledge and process/tool control profoundly. However, the knowledge base is far from adequate at present. Typically, primary chemical reaction paths are not well known, and mechanisms responsible for 3-D microfeature profile determination are certainly a mystery. At the tool scale, fluid flow and conductive/convective heat transfer can be calculated reasonably well, but the impact of specific chemistries, of radiative and transient heat transfer effects, and of plasma excitation of species are poorly understood. These limitations demand a priority on research to develop good 3-D models of chemical and physical behavior on microfeature and tool length scales. In addition, it is essential to determine fundamental reaction paths and kinetic parameters for those reactions exploited in manufacturing and expected to persist through coming technology generations. Finally, processes must be understood not only in the context of 3-D profiles, attendant material and interface quality, and process windows sufficient for manufacturability, but also within the constraints of successful process integration. The sequence of processes employed for microelectronic structure fabrication must be compatible (e.g., sequential etching and deposition to form a particular structure) so they result in the desired material and structure. Process integration imposes constraints on each process as well as the equipment used to carry it out, from geometrical considerations (3-D profiles) to thermal budget aspects (e.g., it is important not to destroy the structure already built by too high a subsequent temperature cycle). 2. Process and Equipment Modeling The need for 3-D understanding and control of processes leads to an imperative for technology modeling at length scales of microelectronic feature sizes, equipment, and perhaps in some cases atomic dimensions. Such modeling promises profound benefit for manufacturing competitiveness in microelectronics because of numerous important uses. Modeling provides a vehicle for optimizing equipment design, and given the intrusion of equipment-dependent complexities on a process, it is essential to focus on equipment modeling. Such equipment models furthermore represent a road map for the evolution of a technology through development into manufacturing, accelerating the rate at which new equipment and processes may be brought into successful operation. Equipment modeling also provides a route for translation of in-situ diagnostic data obtained at a real sensor, into crucial parameters at the point of interest (e.g., at the wafer surface). Modeling of microfeature behavior provi! des an avenue for determining how to achieve the requisite 3-D profiles, what real-time in-situ diagnostics to employ in manufacturing, how to define an optimal set of real experiments in order to define the process and process window, and what are the sensitivities of the process to physical and chemical parameters (thereby identifying which chemical kinetic parameters are most important to obtain experimentally). Process and tool modeling research is particularly beneficial if accompanied by a link to experiments which can be employed to validate and improve the models. Once reasonably accurate models are at hand, the utility of the work is greatly increased if some path is available to make the models available, either through public distribution or commercialization. Either requires a reasonable level of robustness and user-friendliness, but this need does not decrease the premium paid on not simply demonstrating the existence of a predictive model but also transferring it to a broader engineering audience which will use it. This furthermore opens the door to benchmarking exercises which compare the value of different models, and to integration into computer-aided design frameworks which link various related pieces of technology modeling. 3. Equipment While process research focuses on the essential chemistry and physics employed in microelectronics manufacturing, processes only occur within the context of the equipment environment, and the latter complicates both relevant chemistry and physics considerably. Tool scale fluid, thermal, and chemical distributions alter greatly the characteristics of the process as embodied in different positions across a wafer or on different wafers. The surface area consisting of reactor walls can be comparable to that of the wafers in the reactor, so that chemistry at the walls - different for hot wall cf. cold/warm wall reactors -can alter the species arriving at wafer surfaces. Wall reactions, generally different from those on wafer surfaces, can yield products which perform as reactive impurities - even chemical poisons - so that the design of the equipment to embody the desired process becomes at least as important as the identification of the process itself, particularly in the c! ontext of manufacturing effectiveness and competitiveness. Finally, equipment design determines much about the generation of particulates which have a major impact on manufacturing yield, where particulates may be generated by the chemistry/physics on the tool length scale (e.g., gas phase nucleation or ion bombardment in plasma/wall interactions) or by mechanical behavior of moving parts. The enormous cost of manufacturing equipment (approximately 250K/process module at minimum) drives a need to standardize and simplify process equipment. If tool design and process understanding and control could be brought to the point that a given process module (reactor) could be used for several different processes, significant reduction in manufacturing cost may be achieved. The strategies by which different tools are brought together (e.g., integrated process tools, or clusters) has profound potential for defect reduction and reproducibility of critical process sequences, as well as new levels of performance (e.g., enhanced interface properties). Finally, successful equipment utilization for manufacturing requires significant capability in process diagnostics and control, particularly through in-situ sensors and real-time closed loop operation. Identification of effective sensor techniques and optimized control loop strategies are both facilitated by modeling exercises. For manufacturing effectiveness even rather straightforward diagnostics like residual gas analysis, optical emission, etc. can be extremely valuable. In addition, one can anticipate that more sophisticated diagnostics approaches could be developed which are compatible with cost, access, etc. constraints in manufacturing. Taken together, the package of advanced sensors, closed-loop control, and associated response models may promise a kind of generation of "smart tools", with great promise for manufacturing competitiveness. 4. Defects Understanding and control of processes and equipment is an incomplete accomplishment unless the role of both in defect generation is also brought clearly into the picture. Process chemistry itself is already an intrinsic generator of defects. The presence of gas phase reactions, leading to higher intermediates or even particle formation, represents a defect generation path for both deposition and plasma-based etching processes. With high levels of device integration in microelectronics, even extremely low defect levels (ppm or less) can be yield detractors or killers. Another avenue to process-intrinsic defect generation is competing reaction paths. If a reactant set is intended to be used for one reaction pathway, a competing reaction path which occurs at rates 5-6 orders of magnitude lower may still represent a yield killer in manufacturing. This underscores the importance of understanding the process chemistry and physics not only for the intended reaction but also ! for accompanying/competing reactions. Dealing with defects is absolutely essential for manufacturing. At the level of a few defects/cm2 on a wafer, as required for current technologies, defect detection is a major challenge, particularly on patterned wafers. Waiting until device structures are fully built in order to carry out electrical testing for defects is workable but not strategic, since this introduces major time delays and wastes substantial manufacturing costs. Characterization techniques which can identify defects at an early stage after their generation (and possibly on-line removal) would represent great value. Another complication is that techniques capable of discerning submicron size defects (e.g., tunneling microscope, high resolution SEM) may require very long measurement times to "find" isolated defects. Exploiting mechanisms to "decorate" defects is a potentially powerful alternative approach. It is also important in the context of research which seeks to understand the mechanistic behav! ior of defects (e.g., their chemical identity, their chemistry). If defect microchemistry can be revealed, as has occurred in a few cases, then there may be hope of controlling the consequences of defects, such as through chemical processes which may deactivate them electrically. Fundamental studies of defect generation, identification, chemical and physical behavior, passivation, and control have major potential in improving manufacturing. 5. Environmental Aspects In the choice of processes and equipment for increasing levels of thin film processing for microelectronics manufacturing and other applications, concern for safety and environmental impact continue to be an important factor. From an economic point of view, the negative impact of safety hazards and environmental damage are significant and increasing. Therefore a premium is placed on choice of processes and equipment which minimize such risks. Explosive and toxic gases now perform numerous processes in microelectronics manufacturing. Their point-of-use generation, or the utilization of alternative chemical reagents, would decrease safety concerns significantly, and even in the absence of accidents, the cost of safety monitoring equipment and supervision could be reduced. Treatment and disposal of liquids is another portion of this picture. Where wet processing can be replaced by dry, volumes of chemical wastes in need of disposal can be decreased, but substantial process optimization, understanding, control, and associated equipment design are required to make alternative chemistries viable in manufacturing. The consequences of environmentally destructive exhaust gases are also serious, and equipment design should include exhaust treatment techniques which protect the environment. With improved process/equipment modeling and sensor monitoring of microelectronic processes, the potential exists to optimize reagent utilization and the minimization of energy consumption within the constraints of viable product manufacturing. 6. Reliability While product reliability does not have an immediate impact on cost and yield, it plays a crucial role in manufacturing competitiveness because reliability is a necessary condition for quality products (in fact, one often means reliability when one talks about the quality of manufactured products). Microelectronic reliability has been a major technical issue for many years, in such areas as hot electron degradation of MOS structures, electromigration failure in metal interconnections, and structural stability of multilevel interconnects. Establishing empirical patterns for extrapolating stress (temperature, voltage, etc.) acceleration behavior to predict longer-term reliability at normal operating conditions involves major costs and time commitment in the product development cycle. Research could provide major cost savings as well as enhanced product reliability/quality. If new insights were gained into chemical/physical mechanisms which drive significant reliability f! ailure modes, these insights might be converted into process/equipment changes which suppress the failure mechanisms. Also, novel diagnostics techniques capable of detecting incipient reliability failure at an earlier stage would be of profound value, either by reducing the experimental effort required to establish a reliability data base for a technology in development or by providing a reliability screening test for removing product parts which would likely fail too early in the field. Equipment reliability is a very different, but also significant opportunity to enhance competitiveness. Manufacturing efficiency depends sensitively on mean time between failures and mean time for repair of equipment. More robust equipment design as well as enhanced ability to sense incipient equipment failure are worthy goals for manufacturing-related research. REPORT OF THE AUTOMOTIVE INDUSTRY BREAK-OUT GROUP INTRODUCTION The following document describes a list of seventeen items important in manufacturing, with specific relevance to the automotive industries; seven are general in nature and ten are process specific. The automotive industry encompasses almost all known manufacturing processes and while these latter ten items are certainly not the only important technologies that require detailed scientific and engineering research, they seem to be at the forefront of present manufacturing operations. The first seven general items are of critical importance for developing quality, flexibly manufactured products; they must be included as important parameters for developing optimized manufacturing technologies: 1. Simulation and modeling, 2. Optimization of processing through real time measurement and control, 3. Rapid prototyping and short run production methodology, 4. Design for manufacturing, 5. Recycling and design for disassembly, 6. Development of non-polluting processes 7. Plant abatement controls The ten manufacturing technologies considered important for the automotive industry include surface enhancement, metal forming, welding/joining, machining, surface protection and repair, sintering, heat treatment, injection molding, casting and sensor packaging. The panel recommends that all projects funded under a Manufacturing Initiative should be based on multi-disciplinary scientific approaches and should involve manufacturing process and product/material properties interaction. It is also suggested that no proposal funded by NSF in the area of manufacturing sciences should be without an industrial sponsor/co-investigator who will describe in the submission how the proposed research will affect his industry, either through cost-benefit modeling or through some direct applicability to the industry. Industrial collaboration should be with a North American industrial partner and should clearly demonstrate functionality and applicability of the proposed project. A mere letter of support is not sufficient. In addition, the proposal should include a clear statement of how the research will be integrated within the industrial framework of the relevant industry. That is, efforts are also required from the academic institution in the area of technology transfer either through seminars at the industrial workplace or through the students and/or the principal investigator spending some time on the industrial site. NSF review procedures should comprise a panel that includes at least 50% engineers and scientists from industry versed in the product and/or manufacturing process being investigated. BACKGROUND Auto industry product needs in the 21st century will focus on fuel efficiency, light weight, safety, recyclability and pollution control, along with world-class product quality and reliability. Manufacturing requirements will include short run production capability at high first-through quality. There will be a growing competitive requirement for fast new product introduction. Electronic components usage will continue to grow with more customer-valued items as well as more reliance on sensors for product (component) reliability and maintenance diagnostics. OUTLINE (OVERVIEW) OF RESEARCH NEEDS IN MANUFACTURING- AUTOMOTIVE INDUSTRY - not prioritized PROCESSES AFFECTED MATERIALS -PRODUCTS RESEARCH ADVANCES ARE NEEDED IN: 1. Simulation and modeling of manufacturing operations All materials a. CAD/CAE b. Thermophysical database management c. Fuzzy logic d. Methodology incorporating advanced (e.g. parallel) computer architectures e. Stackup statistics 2. On-line process control and optimization. Real time process measurements. All materials a. Design and construction of new sensors (opto-electronic and thermo-acoustic based devices) b. Statistics and tolerancing c. SPC and model-based control theory d. Fuzzy logic e. Knowledge-based systems f. Non-destructive testing g. Database management 3. Rapid prototyping for near net shape and short run production Magnesium, aluminum, ferrous metals, polymers, MMC's and polymer composites a. Heat transfer b. Fluid flow c. Phase stability d. Multi-phase transport e. CAD/Stereolithography 4. Design for manufacturability All materials a. Chemistry b. Metallurgy c. CAD/CAE e. Quality/function/deployment f. Manufacturing science g. Databases on product/material performance versus processing methodologies h. NDT and failure analysis 5. Recycling and design for disassembly Aluminum and magnesium alloys, composites All manufacturing process variables 6. Non-polluting process development Low emission quenchants, coolants, lubricants, etc. a. Process design b. Optimization theory 7. Plant abatement controls Processes affected rather than specific materials a. Gas/particulate collection b. Catalysis 8. Surface enhancement technologies Engineered materials, All metal/non-metal combinations a. Thermal spraying b. Ion implantation c. Laser/electron beam technology d. Plating e. Plasma physics f. Heat transfer g. Surface science h. Tribology 9. Metal forming: rolling, forging, bending, extrusion, stamping Composites, thin coated steels, aluminum, warm-formed magnesium, bimetallics Formability criteria 10. Welding/joining, fastening and assembly Polymer and metal composites, thin coated steels, aluminum, magnesium, bimetallics a. Heat transfer b. Thermo-capillary flow c. Phase stability d. Surface science/materials science e. Mechanical metallurgy f. "Tolerancing Technology" 11. Machining Historic materials plus those listed in 10 above a. Heat transfer b. Surface chemistry c. Mechanics d. Tribology 12. Surface protection (e.g. paint) and repair of damaged surfaces Aluminum, composites, steels a. Surface chemistry/physics b. Corrosion protection c. Spray dynamics d. Materials Science 13. Sintering Powders/sprays of ceramics, metals, engineered materials a. Diffusion b. Surface physics c. Non-Newtonian physics 14. Heat treatment Steels, irons, non-ferrous metals a. Structure control b. Diffusion c. Heat transfer 15. Injection molding Unreinforced and reinforced polymer composites a. Fluid flow b. Reaction kinetics c. Heat transfer d. Tooling 16. Casting: sand, foam, squeeze, die (semi-solid, lo/hi pressure, PM, SPM) Aluminum, magnesium, MMC's, cast irons (gray, nodular, ADI, CG), steels a. Fluid flow b. Heat transfer c. Solidification d. CAD/CAE e. Simulation f. Metallurgy g. Materials science h Ceramics i. Structure/property relationships 17. Sensor packaging Electronic components a. Environmental effects on data 18. Materials handling All materials Design for manufacturing and automation assembly DISCUSSION OF RESEARCH NEEDS AUTOMOTIVE INDUSTRY 1. Simulation and Modeling of Manufacturing Operations One of the major manufacturing issues in the U.S. is the dearth of information about the effect of each process variable on the final product performance (attributes). Competitor countries have large numbers of process engineers who can collect the detailed information necessary for process optimization. For many reasons, this manpower just does not exist in the U.S. However it is impossible to obtain the most robust process at the less product cost without a database relating all the process parameters to the reliable performance of the resultant product in service. Modeling and simulation is essential. Simulation tools are required that can assist the designer in specifying manufacturing processes and equipment as well as assessing the micro-and-macrostructure and ultimately the performance of these materials. Because of the large amounts of data required, it is envisaged that the new generation of massively parallel computers will have to be employed in these progr! ams. This will require developing models that can be run with this new architecture. Modeling will also require verification; thus interaction of model-based programs should be concurrent with sensor-based process measurement and database management, as detailed below. 2. Real Time Process Measurement, Control & Quality for Process Optimization Because of the problems involved in ensuring accurate process control and process optimization methodologies, it is essential to develop accurate collection procedures. As microprocessors continue to be reduced in price with enhanced memory and functionality the development of cost-effective "smart" sensor-based technologies becomes viable even for low value-added, low profit industries. Research programs in this area could be directed to measurement technology using a range of optical, electronic, thermal and acoustic devices mated to the different processing operations in a given technology. However, in addition, there are the methodologies of advanced statistical techniques (e.g. Taguchi, design of experiments) to ensure data quality in terms of signal sensitivity, reproducibility and reliability. This information would then be interwoven with the statistics of the resultant product attributes to ensure that the product user (design engineer, product program manager,! and ultimately the consumer) is able to optimize his or her product design (weight, cost, properties and so forth). 3. Rapid Prototyping, and Short Run Production There is a concerted need to produce prototype components to satisfy the product engineer's needs for an object that can be tested and qualified in as short a time frame as possible. The problem is that conventionally this takes a long time, has a significant cost impact on the car program and may not be representative of the final production component. In order to incorporate new materials in a product, it is important to develop a protocol that produces a prototyped design/product/material with a process that also prototypes the final production process. The product engineer will then be able to qualify his component both as a prototype as well as a final product without the usual high cost and long testing time. The resultant component can then be designed and manufactured simultaneously. Additionally, there is a growing need for tooling that can produce only a small number of components for a short production run. The types of research envisaged would include CAE/s! tereolithography that includes the manufacturing rules (for foundry they could be shrinkage, heat transfer and feeding, for forging they would be by metal flow). Development of low cost dies is required; e.g. by metal spraying, or cast-to-shape, so that new designs can be produced at low cost ($10,000 vs. $100,000) and in a short time (weeks instead of months). 4. Design for Manufacturability The ability to deliver components that are manufactured with the final product attributes in mind is important if the U.S. is to compete with its industrial neighbors. The methodologies will allow the "best" product to be designed simultaneously with a robust process. Product measurements include the entire range of physical, mechanical, chemical and electronic properties; process control attributes include variables that reliably produce various components at a world-class quality (6 sigma), using flexible manufacturing processing techniques. Risk analysis methodology and cost-benefit design criteria are also required. Product and design engineers use existing databases to design components (to near net shape, for crash worthiness and so forth) especially using sophisticated CAD/CAE models. Unfortunately, there are no statistically exact property data available. This requires research to develop a statistically valid database that satisfies all the required environmenta! l conditions to which a vehicle and its components are exposed. 5. Recycling and Design for Disassembly Recycling is a given for every material in the vehicle. This will require some form of bar coding to identify the provenance of each material used so that it can be reconstituted back into its original state. Conventional recycling technology is not appropriate for next generation processing since there is too much mixing and therefore contamination of metallic components. Because of the high demand for near net shape components materials will require increasingly stringent constitutional specifications. New methods for disassembling (and therefore for assembling) a typical vehicle will be required to ensure that each different composition can be easily removed and then recycled. This will allow each component to be re-manufactured without having to reformulate its chemistry. 6. Non-Polluting Process Development The objective is to develop processing methods and materials that reduce the amount of emissions of an existing process. The cost-benefit will be either better product attributes, reduced manufacturing cost (doubtful) or less costly abatement procedures downstream. The type of research envisaged would involve developing new foundry binders and die lubricants that satisfy process/product requirements. 7. Plant Abatement Controls There are may processes (e.g. foundries and heat treatment ovens) that produce large quantities of hazardous and/or toxic emissions. The objective of programs in this area would be to develop methodologies for detection, collection and neutralization of the effluent, and then its safe disposal. 8. Surface Enhancement Technology There are many materials that satisfy most of the attributes of a product but without certain thermal, corrosion-resistant or tribologically important characteristics at one or more surfaces. One research goal would be to apply a coating that does not flake off when applied to soft materials that have large coefficients of expansion. Processing technologies of interest would include thermal spray technology (plasma, transferred arc, high velocity oxy fuels, etc.), and ion implantation, laser and electron beam processing for local surface property enhancement. Even the older processes of electrolytic plating, salt bath nitriding and induction heating require new methodologies and research to ensure world class process control of product attributes. 9. Metal Forming Metal forming is a basic manufacturing technology. Any research that improves the productivity and quality of the final product is important. As new products are designed to meet unusual customer requirements, the ability of the material to flow may not be possible with conventional processing approaches. 10. Welding/Joining/Fastening Assembly Each new material requires an ability to be joined as part of its use in a finished component. This joining may be via a welding or brazing operation, it may also be through gluing or even a Velcro adhesion procedure or through a mechanical fastener and assembly methods. The requirements (process and NDT testing procedures) to ensure a quality assembly that satisfies the component's required attributes (strength, corrosion resistance, etc.) must be defined in order that the material eventually be commercialized. 11. Machining Many of the new materials have a high wear resistance. Since most manufacturing processes require hole-drilling/tapping, milling and so forth, the material's machinability can be a serious bottleneck in the final assembly costs. The ability either to develop a new machining process or cutting tool, or develop cutting tool criteria (speeds/feeds etc.) is important. Machinability problems abound in industry. As productivity requirements have increased and as demands for new cutting fluids have changed, the local tip/workpiece events for machining conventional materials begin to exceed the ability of the tool to remove material while providing the appropriate surface geometric characteristics. An understanding of the phenomena associated with metal removal at high rate, in complex cutting fluid chemistries is required; development of a knowledge-based system, feedback control to develop "smart" system algorithms would improve the quality of the workpiece. 12. Surface Protection and Repair New materials, such as aluminum and reinforced polymer composites, when formed into body panels require for example a protocol for surface repair, either because of damage during production or due to a field problem. In addition, all materials require some surface protection, via for example an electrocoat, chemical conversion coating or painting operation. The methodology of ensuring good adhesion, and a Class A surface quality using a non-polluting procedure is important if the auto industry is to develop new lighter weight, value-added materials. 13. Sintering The ability to engineer powders that can be formed via a series of stamping/sintering operations into a near net shape component has been of interest to automotive engineers for many years. The powder morphological and property requirements and the binder/pressure/temperature envelope for quality, low cost components requires new advances and further development if there is to be an innovation into industry. Sintering is also an important phenomenon as is catalyst functionality, a growing area of importance in the automotive industry. 14. Heat Treatment The properties of almost all metals are affected through a heat treatment. Whether to promote surface hardening or through-hardening to increase strength, or annealing to improve machinability, the processing variables require definition to ensure a robust, economic process. As environmental issues become more important, and as component complexity increases, research is required to ensure manufacturing optimization. 15. Injection Molding There are demands for processes that can produce a high quality polymer and reinforced polymeric components (in terms of surface quality and reproducible properties) at production rates competitive with conventional steel (and even aluminum) stamping body production processing. In addition, issues of part reparability require investigation so that a Class A surface finish be available for final painting. It is critical that the materials ultimately used be recyclable back to their virgin state. 16. Casting Technology The ability to produce a near net shape, high quality cast component is basic to almost all industrial products, from aircraft to automobiles. The choices are myriad varying from plaster, sand and foam on the low solidification rate/low temperature gradient end to high pressure die casting at the high rate/high gradient end of the spectrum. Modern foundry manufacturing issues relate to simulation and modeling and design for manufacturability analyses, to be able to predict cast micro-and-macrostructure and thence the performance of the cast component. The insurance of 6 sigma first-through casting quality, essential to minimizing casting cost, will require detailed control of sand or steel molds, furnace/melting parameters and raw materials using DOE to obtain process robustness. 17. Sensor Packaging There is a growing demand for sensors to enhance a customer's response about his/her vehicle, whether it be through enhanced maintainability, safety or emission controls. Many sensors are exposed to high temperature and corrosion in a complex fatigue environment. The packaging to ensure stable device operating parameters is unique and requires investigation. 18. Materials Handling As automated assembly techniques for short-run, flexible manufacturing systems become more prevalent, techniques for picking, grasping, moving and locating objects become more important. Vision systems and tactile intelligent gripping systems will be required to accurately position assemblies. REPORT OF THE CHEMICAL INDUSTRY BREAK-OUT GROUP INTRODUCTION This report will elaborate on the types of research programs NSF should support in order to improve the economics and thus the competitive edge of the U.S. chemical manufacturing industry. These research programs should have the following objectives: 1. Better control and use of automation 2. Higher quality products 3. More energy efficient processes 4. Enhanced environmental performance 5. More reliable operations 6. Higher productivity o Manpower o Capital o Process of innovation itself 7. Faster implementation time 8. Flexibility The latter two items, faster implementation time and flexibility are what others have referred to as more agile manufacturing. The panel recommends that significant weight be given to the proposals that show meaningful collaboration between the university and a participating company or companies. Such collaboration could take the form of the faculty spending summers at the company; graduate students doing research at the company location; commitment of joint funding from the company; or direct involvement of the industrial scientists or engineers in the conduct of the project. Evidence of such collaboration can take the form of a letter from the participating company and should be included in the proposal. Also, the principal investigator would be expected to prepare a statement of the technical and economic impact of the proposed project on solving important manufacturing problems. OUTLINE (OVERVIEW) OF RESEARCH NEEDS IN MANUFACTURING- CHEMICAL INDUSTRY The following outline of manufacturing research areas are prioritized according to their relative impact on improving the manufacturing enterprise in the chemical and related industries (with the highest priority items listed first, then in descending order). General Research Topics 1. Better Modeling -Databases and better data -Consistency between different level of models -Equation oriented models and use of advanced computer architectures -Dynamic modeling tools -Three dimensional models -More rigorous models versus empirical -Computational fluid dynamics and kinetics -Model solution techniques -Combination of economics and plant design 2. Process Design -Economic optimization -Process synthesis o Flexibility o Waste Minimization o Controllability o Safety o Maintainability o Constructability o Flexibility o Energy integration o Separation sequencing -Handling of uncertainty o Decision and risk analysis -Batch Process Design and Scheduling 3. Reaction Engineering -Mixing models -Computational fluid dynamics -Mechanistic Reaction kinetics -Rigorous Global Models (for example, Combustion, -Polymerization, Reaction injection molding, Packed-bed catalytic reactors) o Heat transfer o Mass transfer o Momentum o Chemical Kinetics -Multiphase systems -Molecular reactor control (control every molecular interaction) -Catalysts o Robustness o Deactivation o Reactivation -Combustion 4. Affordable Sensors and Analyzers -Sampling Techniques -Composition analyzers -On-stream -Quick response or continuous -High Temperature and pressure -Advanced statistical techniques 5. Better Control Strategies -Human interactions -Fuzzy logic -Knowledge based control -Model based control (multivariable, nonlinear) -Data reconciliation (model-based, statistical) -Use of advanced statistical techniques -Improved dynamic simulation tools o Validation o Training 6. Particle Technology: formation, control and handling -Control -Size, shape, structure -Crystallization -Grinding -Precipitation -Particle transport and storage o Fundamental models needed -Solid fluid separation -Particle drying 7. Polymer processing -Monomer solubility -Computational fluid dynamics -Devolatization -Extrusion -Mixing of widely different viscosities -Constitutive relationships 8. Separation -Field enhanced separations -Rate based separations (membranes; PSA) -Dynamic simulation 9. Product Design (Need to be able to go both ways) -Molecular modeling -Microstructure -Macrostructure -Bulk physical properties -Product performance 10. Process Intensification -High gravity distillation -Ultrasound -Turbulators -Transfer is resistance times area times driving force. Process intensification will address hanging one or more of these three items) DISCUSSION OF RESEARCH TOPICS/NEEDS CHEMICAL INDUSTRY All research opportunities have to have a common goal and metrics for those goals to ensure that one achieves the concerted coordinated effort one is seeking. For example, some chemical companies have developed mechanisms that characterize manufacturing with certain attibutes with attendant metrics as follows: ATTRIBUTE IDEAL METRIC First pass, first quality yield 100% Cycle time 0 Process capability Greater than 2.0 Up-time 100% Defects 0 Transition time 0 These metrics are being used for indutries such as bulk chemicals, fibers and electronics. The following specific areas need to be considered as part of such a coordinated effort. 1. Mathematical Process Modeling Validated, accurate mathematical models for a variety of chemical processes are crucial in process analysis, simulation, control, optimization, and design. In order to contruct effective models, it is important to have comprehensive databases of relevant parameters such as physical properties, hence the continued upgrading of databases for the chemical industry is a high priority. In dealing with a range of models that require different levels of microscopic detail, it is desirable to have consistency between the models at each level. There should be a uniform protocol and consistent architecture so that various levels of the models can communicate effectively with each other. Software packages with equation-oriented models, which are amenable to programming in advanced computer architectures, are preferable. These equation-oriented models utilize sets of differential equations rather than procedural models. The development of dynamic modeling tools and models with three-dimensional capabilities are becoming feasible with increased computational speed in new machines (such as with parallel processing). The use of rigorous models based on fundamental chemical and physical phenomena is preferred over empirical models. This will permit extensions and applications of the models beyond the original database. There is a need to link advanced computational fluid dynamic models that are now available commercially with chemical kinetic models. In addition, new mathematical techniques that speed up the solution of complex models need to be developed. Finally, the meshing of plant design models and economic models are necessary in order to arrive at optimum total plant designs. 2. Process Design The objective of process design should be the economic optimization of the supply chain (supplies --- plant --- customer). It is important that optimization of the plant not be done in isolation because this leads to sub-optimization of the supply chain, leading to excess inventories at the supplier, distributor, or customer's location. One of the earliest decisions that needs to be routinely made is whether the plant should to be designed for batch processing or continuous processing operations. Subsequent steps include the design of individual unit operations. The overall integration with marketing models should be the ultimate objective of chemical process design. Computer software that optimizes design and scheduling is therefore desirable. An important step in design is process synthesis. This entails configuration of the optimum unit operations in order to achieve flexibility, waste minimization, controllability, safety, maintainability, constructability, and energy integration. This step has a major impact on the economic viability of a process or plant, and comprehensive design tools that deal with these multiple objectives are greatly needed. In addition, research on the effective handling of uncertainty is important in selecting the processes to be used. Decisions and risk analysis techniques using variables that are not fully quantifiable would be beneficial. 3. Reaction engineering Mixing or residence time models are extremely valuable in detailed reactor designs. Computational fluid dynamic programs are beginning to be applied to reactor design and also have broad applicability to the study of flow in pipelines, heat exchangers and even in crystallizers. The inclusion of mechanistic (fundamental) reaction kinetics in reactor models is encouraged for both primary reactions and by-product reactions. It is important to model each step in the reaction sequence in order to optimize the yield of the desired product and minimize the yield of by-products and waste. Whenever possible, rigorous global reactor models, i.e., models that include heat transfer, mass transfer, momentum transfer and chemical kinetics, should be employed. Global models could be extremely useful in designing combustion systems, polymerization reactors, reaction injection molding, and packed-bed catalytic reactors. Improved understanding of multi-phase systems, including two an! d even three-phase systems would also be valuable for reactor modeling. Molecular reaction control, in which every single molecular interaction can be manipulated, is a desirable objective of any chemical process. If one can control reactions at a molecular level, it would be possible to control precisely the production of the amount of desirable product and simultaneously minimize the production of undesirable by-products. In the area of catalysis, the panel recommends increased emphasis on imprving the robustness of catalysts, understanding and retarding the deactivation of catalysts, and more effective regeneration of catalysts. The panel also encourages additional research on combustion, particularly the combustion of organic materials that may be by-products or waste from chemical processing. This research should focus on the chemistry/kinetics/temperature/mixing in the combustion zone (which should be treated like a chemical reactor). 4. Process Sensors and Analyzers The objective of this program should be affordable, reliable, on-stream, quick-response, analytical techniques that can be installed in the production process. The development of accurate sensors that determine composition, and remote detection devices should be especially emphasized. New and novel measuring techniques need to be developed because, for example, it is often desirable to know what is going on inside an extruder in enough detail to analyze the temperature and shear history of an element. High tech devices such as LIDAR to sense chemical composition in a reactor volume element in a non-intrusive manner or from some remote location could have far reaching benefits. In addition to the activity in the development of new sensors, specific emphasis on high temperature and high pressure sensors is also needed. Often the most difficult problem in process analysis is determining the sampling techniques to use. Research into new methodology for sampling and the handling of experimentally obtained, industrial data would be worthwhile. Work on advanced statistical techniques to determine the reliability and stability of the sampling instruments and the accuracy of the data would also be worthwhile. 5. Process control strategies Process control strategies need to be incorporated into process synthesis methodologies. The control strategies to be implemented depend to a large extent on the specific process involved. For example, model-based control is particularly useful for multivariable, nonlinear processes for which single loop control is inadequate. One of the important considerations with any control strategy is data reconciliation, namely what data whould be considered, how accurate are the data, and how the control system should respond, based on those data. The adaptation of advanced statistical techniques to validate and evaluate data sets used as inputs to control algorithms could have far-reaching consequences in the chemical process industries. An important element of any control strategy is the man/machine interface. Development of ways to prioritize the available information and to facilitate human interaction would be worthwhile. This is an area where fuzzy logic could be effective because not all process variables can be rigorously quantified and human or instumental action must be taken based on incomplete data sets. The use of dynamic simulation tools for the training operations personnel would be desirable. In addition, new ideas on how to validate a dynamic model before it is installed in the plant control system are needed. 6. Particle technology The production, control, handling, and storage of particles is a major challenge in virtually every chemical process plant. Techniques based on fundamental modeling to predict size, shape and structure of particles would be extremely valuable in unit operations such as crystallization, grinding and precipitation, because existing models to predict how particles behave in transport and storage are inadequate. Rigorous particle drying models which can be scaled up to commercial size would be very useful. Improved predictive tools for solid/fluid separation, both solid/liquid and solid/gas are also needed. 7. Polymer Processing Computational fluid dynamic methodology should be applied to the various stages of polymer processing (polymer extrusion and mixing), particularly for materials with widely differing viscosities. Further scientific investigations into the fundamentals of devolatization and additional data for very dilute monomer/polymer solutions are needed 8. Separation Many of the conventional separations such as distillation, are quite well understood. Additional work on field-enhanced separations and rate-based separations (such as membranes and pressure swing absorption) is needed. Experimentally validated dynamic simulators for separation processes are needed as tools for some of the other modeling applications discussed earlier. 9. Product design The ultimate aim of product design in the chemical industry is to develop a methodology to relate customer satisfaction to the desired product performance (price, specifications, delivery, and service), then relate product performance to the bulk physical properties of the material, which in turn is related to the microstructure of the material and understanding of how the process affects the desired properties. It is important that this understanding enables one to move readily from the molecular model to the finished part and visa versa. Ideally methodologies will exist in the future that will enable scientists and engineers to specify some product performance criterion from which it is possible to develop various molecular models of materials that would meet those desired product specifications. 10. Process Intensification The rate of transport in a process is proportional to the product of three terms: area multiplied by driving force multiplied by resistance to flow. This particular relationship is the basis for sizing essentially all equipment in a chemical plant. Often the engineer accepts as given, one or more of these factors and then selects the most appropriate equipment. Research to enhance any of these three factors (called "process intensification") could have significant impacts on plant capital requirements. Commercial examples include high gravity distillation, ultrasound to achieve either increased area or increased driving force, and turbulators to reduce resistance to heat transfer. Similar novel ideas should be encouraged. EXTEMPORANEOUS THOUGHTS Research is needed in many areas that are supported by NSF, though not specifically by the CTS Division. These areas include: 1. Rotating equipment improvements 2. Materials of construction o Understanding the corrosion mechanism. o Understanding the effect of trace components in the material. o Understanding the protective coating. o Understanding the effect of impurities in the process stream. o Understanding the effect of the physical environment. 3. Electrical reliability 4. Ceramics processing. CONCLUDING REMARKS Competitive manufacturing requires intelligent efficient production procedures in everything from the individual process reactor to the organization of the factory floor. Some attributes of competitive manufacturing include: (1) rapid time to market, (2) rapid turn-around time for learning, (3) rapid prototyping, (4) efficiency, (5) manufacturability, established at early stages of development, (6) reliable production (equipment reliability, repairabilty, etc.), and (7) product reliability. The chart on page 41 is a visualization of how various factors interrelate to achieve competitive manufacturing. Within this framework of competitive manufacturing there is an interdependence of "manufacturing processes (MP)" and "tools for intelligent manufacturing (TIM)". Manufacturing Processes (MP) vary from industry to industry but there is a commonality between industries. Improving manufacturing competitiveness requires improvements in the specifics of processes and equipment. This is the classical domain of the "hardware-focussed" R&D enterprises. Examples of these manufacturing processes include, for the three industries represented at this workshop: 1. Microelectronics Industry - chemical kinetics, plasma etching, chemical vapor deposition, equipment design, process sensors usable in-situ in real time, etc. 2. Automotive Industry - surface enhancement, metal forming, machining, welding, etc. 3. Chemical Industry - reaction kinetics, multiphase flow, catalysis, polymer engineering, separations, etc. At the same time, real leverage toward competitive manufacturing is engendered by the application of "intelligent manufacturing" to the specific process and equipment (hardware) challenges of a manufacturing industry. Intelligent manufacturing includes process and equipment modeling, sensor applications, model-based process control, and process design. A conclusion of the workshop was that real leverage in MP resides in part in the advantages gained from integration with TIM. Were industries not in the business of making "things" (such as automobile fenders and computer chips) and "stuff" (for example pure chemicals), there would be no MP and therefore no motivation to pursue the various elements of TIM discussed above. Simultaneously, application of TIM elements not only adds value to MP but also defines how to optimally pursue MP improvements in hardware. By carrying out process modeling exercises to determine sensitivity of a process to several physical or chemical parameters, one can identify which experimental work is most important and deserving of attention to directly enhance the hardware aspects in MP. In addition, the value of TIM elements can only be assessed when tested in the context of specific applications in MP. These concepts are depicted in the chart on page 42. As the present organizational structure in the Engineering Directorate of the National Science Foundation is constituted, MP is located in the Chemical and Thermal Systems Division (CTS) while TIM is in the Design and Manufacturing Systems Division (DDM). The above observations then indicate that DDM and CTS are highly interrelated within the context of a manufacturing initiative. Therefore for any research direction related to CTS activities, an added value can be derived from close collaboration with DDM. In fact neither individual piece of the picture can function optimally in a vacuum - in other words TIM and MP are complementary and both parts are needed to make a complete whole. In terms of proposal evaluation by the NSF, proposals which recognize and implement both pieces of the puzzle should have the highest priority. RESEARCH OPPORTUNITIES IN MANUFACTURING IN THE PROCESS INDUSTRIES WORKSHOP National Science Foundation 1800 G Street, N.W. Washington, DC 20550 Room 543 December 2-4, 1992 Wednesday, December 2, 1992 8:30 a.m. Introduction and Goals for the Workshop (Burka) 8:35 a.m. FCCSET Intitiative in Manufacturing (Bordogna) 9:05 a.m. Intelligent Manufacturing Systems (Hodgson) 9:20 a.m. Research Emphases of CTS (McGee) Manufacturing Related Research in the Engineering Research Centers (Lih) 9:45 a.m. Views of Government's Role in Manufacturing (Wong, Pfeiffer, Abramowitz) 10:30 a.m. Break 10:45 a.m. Opportunities/Needs in the Metals Industry (Richmond) 11:10 a.m. Opportunities/Needs in the Automobile Industry (Cole, Mazumder) 12:00 noon Lunch (catered) 1:00 p.m. Opportunities/Needs in the Microelectronics Industry (Ibbotson, Geyling, Rubloff) 2:15 p.m. Opportunities/Needs in the Chemical Industry (Varilek, Siirola, Snyder, Sciance, Edgar) 3:00 p.m. Break 3:15 p.m. Continuation of Chemical Industry Presentations 4:30 p.m. Discussion of Break-out Groups and Assignments Dinner - Dominique Restaurant 1900 Pennsylvania Avenue, N.W. (each Break-out Group to dine together) Thursday, December 3, 1992 8:30 a.m. Three parallel discussion sessions 12:00 noon Lunch (catered) 1:00 p.m. Three parallel discussion sessions (Preparation of draft report) Friday, December 4, 1992 8:30 a.m. Summary of Research Issues - Group I (Discussion) 10:00 a.m. Summary of Research Issues - Group II (Discussion) 11:30 a.m. Summary of Research Issues - Group III (Discussion) 12:30 p.m. Lunch 1:30 p.m. Continuation of Group III 2:30 p.m. Writing Sessions PARTICIPANTS FOR MANUFACTURING WORKSHOP DECEMBER 2nd, 3rd & 4th, 1992 Dr. Stanley Abramowitz National Institute of Standards and Technology Building 222 Room A353 Gaithersburg, MD 20899 Dr. Gerald Cole Ford Motor Company Research Laboratory MD 3182, P.O. Box 2053 Dearborne, MI 48121 (313) 322-1860 FAX# (313) 594-4929 Dr. Thomas Edgar Department of Chemical Engineering University of Texas at Austin Austin, TX 78712-1062 (512) 471-3080 Dr. Franz Geyling Sematech 2706 Montopolis Drive Austin, TX 78741 (512) 356-3014 FAX# (512) 356-3521 Dr. Thom Hodgson Division of Design and Manufacturing Systems Room 1128 National Science Foundation Washington, DC 20550 (202) 357-9508 Dr. Dale E. Ibbotson AT&T Bell Laboratories Room 6F-317 600 Mountain Avenue Murray Hill, NJ 07974 (908) 582-2888 FAX# (908) 582-2913 Dr. Marshall Lih Division of Engineering Education and Centers Room 1121 National Science Foundation Washington, DC 20550 (202) 357-9707 Dr. Jyoti Mazumder Department of Mechanical Engineering University of Illinois Urbana, IL 61801 (217) 333-1964 Mr. John Pfeiffer Assistant to Department Associate Director Nuclear Energy Branch Office of Management & Budget 725 17th Street, N.W., Room 8002 Washington, DC 20503 (202) 395-3702 Dr. Owen Richmond Alcoa Technical Center 100 Technical Center Alcoa Center, PA 15069 (412) 337-2998 Dr. Gary Rubloff IBM Research P.O. Box 218 Yorktown Heights, NY 10598 (914) 945-1142 Dr. Thomas Sciance Conoco Inc. 100 South Pine Ponca City, OK 74603 (405) 767-4716 FAX# (405) 767-4014 Dr. Stan Settles Office of Science and Technology Policy New Executive Office Building Washington, DC 20506 Dr. Jeffrey J. Siirola Research Laboratories B95 Eastman Chemical Company Kingsport, TN 37662 (615) 229-3069 Mr. Irving Snyder Director, Process Technology Development Dow Chemical Company 220 Building Midland, MI 48640 (517) 636-0266 Mr. Randall E. Varilek Director of Manufacturing Technology Air Products & Chemicals 7201 Hamilton Boulevard Allantown, PA 18195-1501 (215) 481-4002 Dr. Karl Willenbrock Technology Administration Herbert Hoover Building, Room 4841 Department of Commerce Washington, DC 20230 POSSIBLE ACTIVITIES OF THE CHEMICAL AND THERMAL SYSTEMS DIVISION OF NSF AND A STRATEGY TO RESPOND TO THE NSF MANUFACTURING INITIATIVE Gerald S. Cole Ford Motor Company, Research Laboratory It is apparent that the United States has a major problem competing with the world manufacturing giants of Germany and Japan. While there are many who wish to point to our growing export markets, these are not in the important area of basic manufacturing "capital goods" products. Manufacturing industries are central to long-term productivity and economic growth of every country. As we look at our competition, we can see that there is a growing battleground for world markets, which the United States is losing. It is essential that the manufacturing initiative of the NSF is begun as soon as possible. There are so many problems that it is difficult to know where to start. One important observation is that young people are not motivated to enter manufacturing as a career; all the smart aggressive students seem to be driven to business, medicine, law, and even other engineering professions but not manufacturing. Manufacturing excellence requires an enormous database which relates how management, the workforce and machines convert raw materials into finished products. Given this dearth of talent, we have to cope with the fact that there just is not enough of the "right stuff" available. So without human resources, how are we to obtain this data, and how are we to do this in a fashion that allows rapid response to competitive thrusts in areas of processing new products and materials? The answer lies in computerizing the processes by which manufacturing controls its output. When we look at say the automotive industry, it is evident that there is a great deal of computer modeling and simulation being performed, but only on the product and components used in the vehicles. There is little effort devoted to simulating the manufacturing operations used to make these components. If NSF devoted its human, institutional and financial energies to develop software and hardware that generate processing-related databases, we could leap ahead of the successful manufacturing countries, all of whom have an enormous resource of human talent, collecting, analyzing and optimizing the critical factors affecting productive output. Since we do not have the people we must look for ways that can simulate manufacturing operations. Most products are produced by processes that have thermal energy associated with the final product somewhere .... casting, forging, injection molding, heat treating, machining, and so forth. It is for this reason that the CTS Division has an opportunity to address manufacturing from a phenomenological point of view in terms of understanding and control using simulation. Since agile, flexible manufacturing requires an even more substantial understanding of the characteristics that affect product quality and productivity, the opportunity for a successful impact on our country's competitiveness is even more striking. It is important that the U.S. begins to cope with world-wide competition and excellence in rapidly producing new products, and making profit at it. In addition to the technical issues that the CTS Division must address in the area of simulation and modeling of manufacturing processes, there are issues of management and workforce training. NSF's historic role in education will be required to nucleate and foster the new methodology in academe and industry. RESEARCH OPPORTUNITIES IN AGILE MANUFACTURING Thomas F. Edgar Department of Chemical Engineering University of Texas Austin, TX 78712 The U.S. chemical industry is dominated by large-scale continuous processes, with the notable exceptions being the food, pharmaceuticals, and specialty chemical sectors. While commodity chemicals will continue to be important, the next 20 years will see more emphasis on rapid delivery of smaller quantities of differentiated products ("time-based factories"). Plants will become smaller and located more closely to customers, with increased usage of batch and semi-batch configurations to shorten response times and reduce inventories. These plants will need to be more flexible (agile) in operation and must satisfy stringent safety, health, and environmental regulations. These smaller plants will require a high degree of automation in order to reduce manpower costs and maximize productivity. This in effect will yield capacity increases per unit of investment. The purview of chemical engineering technology in agile manufacturing will be broad-based, covering advanced mate! rials processing (including microelectronics), bioprocessing, and specialty chemicals (especially polymers). Unit operations will become more diverse, especially in plastics, ceramics, pulp and paper, and microelectronics, which will require development of new design principles. Traditional unit operations may require some new emphases; e.g., for distillation columns, more studies on packed columns and batch systems are needed. Batch processes exhibit more pronounced nonlinearities than their continuous counterparts and thus demand more efforts in the process modeling and control areas. One example where quantum change in processing philosophy is likely to occur is in the microelectronics industry. Presently the dominant technology for thin film deposition is the multiwafer low pressure chemical vapor deposition (CVD) system, which suffers from yield and process control problems. An emerging alternative (although not yet commercially viable) is single-wafer processing using rapid thermal CVD; by operating at higher temperatures it is possible to process larger wafers individually at a function of the time for the multi-wafer system. Many types of deposition reactors will be required, although it may be feasible to develop a flexible, multi-use reactor which could carry out novel processing techniques, e.g., silicon deposition, oxidation, and nitridation performed sequentially in the same reactor. The success of this approach will hinge on developing fundamental mathematical models for process equipment, optimized design, new real-time sensors, and ad! vanced process control strategies, all of which will require significant basic research efforts. There is also a need to develop a holistic approach to design and operation of the microelectronics "factory of the future', including process synthesis, waste minimization, automation, and unit and products scheduling (tracking of individual wafers). In this application there is the opportunity to introduce operational issues (dynamics and control) in the process design step, but this is a murky area even for traditional chemical plants with no clear set of principles to follow. Increased use of batch processing demands the development of advanced scheduling tools in order to maximize productivity. While the field of operations research has addressed some important issues in discrete parts manufacturing, analogous problems in batch processing for chemicals manufacture are quite different and cannot be solved efficiently by traditional OR scheduling/optimization techniques. The complexity of scheduling problems in pharmaceuticals, specialty chemicals, and pulp and paper plants is influenced greatly by equipment out of service, and other unplanned events; "reactive" scheduling, which might be analogous to feedback control in the continuous industries, is a subject that is just beginning to receive attention by chemical engineers in the operations research community. The advent of pipeless batch plants (use of flexible hoses) and moveable batch equipment (reactors on wheels) are two developments that have principally occurred outside the U.S. but! have significant ramifications for improved design and scheduling of process operations. SOME THOUGHTS ON THE NSF MANUFACTURING INITIATIVE Franz T. Geyling SEMATECH It will be a pleasure to participate in the NSF workshop on the title subject. Indeed, the need for versatile, automated and programmable manufacturing techniques is a recurrent theme in U.S. industry these days. It carries strong innuendoes of material science (research) and processing disciplines (equipment/control technology). In fact, the ultimate success of this endeavour will depend on a productive mix of the two. In terms of infrastructures, the challenge is to synergize the necessary RDMC (Research-Development-Manufacturing-Commercialization) cycles. The cultural shift in these collaborations must be toward higher levels of risk in return for potential breakthroughs. Specifically, the early stages of most ventures must be targeting (dis)proofs of feasibility for innovative ideas. The NSF initiative will be a welcome reinforcement of seminal efforts in several manufacturing industries. In the silicon IC area, one can envision collaborations between the NSF and organizations like the Semiconductor Industry Association (SIA), SEMATECH et al. A particularly promising technology is advanced single wafer processing (ASW), which sets the stage for a continuing increase in wafer size (targeting 12" soon after 2000) and the need for elaborate and sophisticated processing of single wafers with sufficiently high throughput. It encompasses low pressure chemical vapor deposition (LPCVD), rapid thermal processing (RTP) and plasma enhanced processing (PEP). The industry will probably search for a somewhat standardized, compact reactor architecture that can be fitted with various process-specific features; e.g. automated gas injectors for chemical cleaning, etching and deposition of films, rapid heating and cooling, and IR, UV, plasma, neutral-beam assists. Wafers will be passed through such diverse modules by a robot handler in a 'cluster' environment. This manufacturing environment can be extended further by including spin-coating modules with metered resist dispensers and in-situ post bake facilities. Looking eve! n further, single wafer fabrication from bulk silicon may become part of this integrated processing/equipment scenario. Finally, chemical kinetics modeling, which is a concomitant of many of the above process studies (LPCVD, RTP, PEP, etc.) must be pursued as a broad-based effort. (I plan to support these remarks with several SEMATECH planning foils that illustrate the anticipated networking between program coordinators, academia, national labs, industrial consultants, equipment vendors and IC manufacturers). Similar perspectives of innovative team efforts can be developed for other technologies, such as bulk (semiconductor) crystal growth and photonics (e.g. pre-stressed multi-fiber assemblies, ultra-rapid coating, and continuous manufacture from glass materials to coated fiber.) RESEARCH NEEDS IN MANUFACTURING SEMICONDUCTOR TECHNOLOGY Dale E. Ibbotson AT&T As the dimensions of the newest and future electronic devices shrink further into the deep submicron range, many of the current processes used today will not be able to meet the manufacturing requirements for tomorrow's factories. The success of any company in this business will depend on the ability to bring new devices to the marketplace with great efficiency while avoiding the extremely high cost of a new fab line. The manufacturing tools and the cleanroom infrastructure required in future semiconductor fabrication lines will become almost prohibitively expensive according to today's standards, unless we are able to turn around the trend in cost versus performance of tools and the integration of processes to produce products. In has recently been reaffirmed that in order to meet the challenges of making the next several generations of devices in a cost-competitive manner, process modeling and equipment design are increasingly important. It is also apparent that in order to model technologically important processes and to codesign process equipment, a knowledge base needs to be available for use in developing models and CAD tools. The knowledge base comprises chemical and physical data on reactions, experimental kinetics, theoretical calculations to obtain rate constants, cross sections, yields of reactions, etc. Until we are able to supply this knowledge base on technologically important systems and use these to design reactors for semiconductor fabrication, the technology and efficiency of manufacturing will suffer or will not be possible. I will explain this particular approach to meet future manufacturing needs. THE ROLE OF CHEMICAL AND THERMAL SCIENCES IN MANUFACTURING Jyoti Mazumder Department of Mechanical Engineering University of Illinois at Urbana-Champaign Urbana, IL 61801 A great number of manufacturing processes are driven by temperature cycles experienced by the materials. Fundamental understanding of the associated transport phenomena (mass, momentum and energy) of these processes is crucial for the development of a quantitative relationship between the manufacturing process parameters and their dependent variables such as mechanical and physical properties, since it is the essential ingredient for process control and automation for smart machines. Theoretical research for predictive capability complemented by strong experimental effort on on-line process diagnostics will accelerate the eventual development of smart machines and agile manufacturing. In order to achieve the above mentioned quantitative relationships, challenges and opportunities are in the cross-disciplinary research between materials science, physics and thermal science. It may appear from a general perspective that many of the associated transport phenomena are already well-known and well-understood. But a closer scrutiny will reveal that cross-disciplinary interaction between materials science and thermal science is at its infancy. This is imperative for predicting the properties which are based on microstructure resulting from the thermal cycle. For example, even for well-established processes such as heat treatment, conduction heat transfer may provide the temperature profile but the hardness information is only obtained when heat flow data is used with hardenability data and time-temperature transformation (T-T-T) diagrams. It has been known for some time that nucleation and growth controls the T-T-T but still much is left to be desired to develop the quantitative predictive capability. This type of knowledge gap becomes even more acute when one ventures into processes involving solidification, such as welding, cladding and allo! ying. Until the recent past, even the prediction of the solid-liquid interface was a difficult proposition due to the lack of understanding of the thermocapillary flow in these processes. Complexity of microstructure prediction increases due to the non-equilibrium partitioning at the solid-liquid interface. It is only recently that such problems have been addressed. Integration of heat and mass transfer models with the introduction of non-equilibrium partition coefficients derived from atom trapping theory as a boundary condition at the solid-liquid interface lead to the capability of predicting non-equilibrium phase diagrams. Knowledge derived from such a non-equilibrium synthesis technique is successfully applied to develop a process to clad an aluminum engine head of an automobile. A high power laser was used for this cladding process for controlled energy input. The fruit of this process is a light engine with higher operating temperature due to the seven-fold incre! ase in thermal diffusivity at the valve seat compared to what is offered by the present technology. This in turn will provide higher torque for a smaller energy efficient engine. Automobiles with components using this technology are expected to be available in the market place by 1994. Another facet of cross-disciplinary thermal science research are the challenges and opportunities provided by specific emerging technologies such as laser processing. For example, plasma produced during the laser processing lead to refraction and absorption of the incoming energy source and thus introduces uncertainty in the fundamental boundary condition of this form of thermal processing. Yet an effort to solve this problem is still in its infancy. There are several interesting thermal science issues needing attention, such as time (femptosecond) and intensity (up to terrawatts per square centimeter) dependent absorption; turbulent mixing of expanding plasma and vapor with the incoming shielding gas used for protection from oxidation; intensity dependent ionization and recombination, discontinuity at the liquid-solid interface and vaporization and recondensation in the laser induced cavity or "keyhole". Any contribution in these topics will lead to a robust manufact! uring process. Development of the diagnostic techniques for the model verification and on-line monitoring of the important process parameters is another cornerstone for eventual development of smart machines. This area, along with plasma related issues, will need close interaction of thermal science and physics. Some important achievements can be cited in this regard for combustion research and those can be explored for controlling processes such as welding and chemical vapor deposition. Manufacturing processes such as heat treatment, casting, welding, cutting and surface modification affect almost all materials in an automobile where more than 45% of the cost is materials. Considering that one job out of every seven is automobile-related and the present day automobile is one of the largest users of electronics, one can conclude that improvement of the manufacturing processes related to this sector will enhance national competitiveness. SOME RESEARCH NEEDS AND OPPORTUNITIES RELATIVE TO THERMAL AND CHEMICAL ASPECTS OF METAL MANUFACTURING AND APPLICATIONS Owen Richmond Alcoa Technical Center Alcoa Center, PA The objective of metal product manufacturing is to achieve products which satisfy desired performance capabilities. These capabilities depend upon certain attributes of the product such as geometry, residual stresses, internal "structure" and "surface" structure, which in turn depend upon a whole chain of processes. Increasing traditional empirical methods of process development are being complemented by the use of physically-realistic mathematical simulations which reduce development time and cost and need to improve the product quality and process efficiency. These simulations form the basis for concurrent design of products and their manufacturing processes and also for the optimization of the entire system of unit processes. Some examples will be taken from the aluminum products industry with particular emphasis in the research opportunities related to the thermal and chemical aspects of both manufacturing processes and product applications. RESEARCH NEEDS IN MANUFACTURING Gary W. Rubloff, IBM Research The national initiatives on manufacturing bring forth challenging questions beyond the usual domain of our research enterprise. Let me convey some thoughts for discussion, no doubt reflective of my own involvements in microelectronics science, technology, and manufacturing research. Research role in manufacturing ---- manufacturability Research needs in manufacturing should be just that - opportunities for science and applied research to impact, not do, U.S. manufacturing. The NSF research constituency can improve manufacturing by redirection to research areas which are as intriguing and challenging but at the same time focussed on issues pivotal to manufacturing. In any case, this means concentrating on understanding the science and on controlling the physical/chemical phenomena (the engineering part) well enough to predict and optimize manufacturability; this requires a reasonable understanding of what is important in manufacturing, but is quite different from working on or improving manufacturing. Another view of this distinction is that the research community's value to manufacturing competitiveness is at a point of higher leverage, such as providing more effective materials and processes, equipment, designs and design approaches, operational controls, etc., all of them stemming from research issues. Interdisciplinary demands Manufacturability and efficient manufacturing require an incredibly interdisciplinary expertise. For example, many of the important technical components for chemical vapor deposition of metals are included in NSF's Chemical and Thermal Systems area - kinetics, reaction engineering, heat transfer, process design and control, but manufacturing application forces much closer ties between these areas. Equipment design, gas flow, and heat transfer alter chemical reaction rates throughout the reactor, while the energetics of specific reactions further modifies heat distributions and gas flows. The presence of reactor walls, often at different temperatures, provides additional reaction products to compete with primary, intended reaction paths. And in most important cases, the primary reaction pathways and kinetics are either unknown, or at least poorly known. Efficient manufacturing necessitates rapid turn-around-time (TAT) for learning, suggesting in-situ diagnostics/real-time process control and technology modeling - directions requiring still other areas of expertise. And finally, the technical world - from research, development, and manufacturing - changes so rapidly now that its contributors must be capable of quick change of focus from one type of problem! to another, necessitating strong interdisciplinary awareness or talent. Equipment for manufacturing effectiveness Microelectronics manufacturing has historically separated process and process/device integration from the specifics of the equipment (or tooling). Effective manufacturing requires a reversal of this attitude. From reactor design to chemistry on reactor walls to the chemical engineering of reactant delivery systems, the equipment details are as important as the intended process, capable of dominating defect densities (e.g., spurious wall reactions producing unwanted contaminants), and these details also complicate flow patterns, thermal and process uniformity, etc. Equipment design also affects the efficacy of in-situ diagnostics and real-time process control strategies. Relation between equipment-scale behavior and product performance While equipment/reactor design and process choice are the primary determinants of chemical and thermal environment which generates product (e.g., semiconductor wafer, thin film coating), the chemistry and physics presented at the product surface has a subtle but crucial connection to product quality. In thin film processes generally, this means that the film properties on the product depend on local chemistry and physics in ways which must be understood (e.g., to relate local process parameters to electrical or other properties of the films). In 3-dimensional microelectronics, the submicron-scale shapes of features (e.g., deposition or etching profiles) are equally critical to microelectronic performance and are determined by local chemistry and physics. From the point of view of chemical and thermal systems, this means that one must understand: the operative local chemistry/physics, the consequences on local product properties, the way in which equipment/tool configuration deliver reactant and thermal profiles to the product, and how process/tool diagnostics can assure the desired and necessary conditions in the tool. Stated another way, reactor engineering design (chemistry and thermal) is inextricably linked to the consequences for the manufacturing product on both microscopic and macroscopic scales, as well as to process/tool control strategies. Virtual manufacturing - the role of technology modeling Models for technology - equipment, process, and integration - offer enormous promise for leading-edge manufacturing. Early learning about equipment and process design save capital investment costs and shorten TAT to develop processes and define tools. Models also provide road maps for rapid solution of problems from development to manufacturing bringup and crisis recovery in steady-state manufacturing. Even where models are rudimentary (e.g., due to incomplete knowledge of chemical kinetics) they can be valuable, such as in relating figures of merit to process parameters, identifying process diagnostic strategies, relating in-situ diagnostic measurements to critical parameters, and determining parameter sensitivities to drive further research. At the opposite extreme, operational models have profound economic benefit, from factory floor layout to cost-of-ownership. Simulation and visualization Equipment and process design to enhance manufacturability and manufacturing efficiency can be optimized at an early stage if simulation of equipment and process can be carried out. For example, the equipment designer benefits by "running" virtual equipment to see if it performs well before actually building real equipment. The process designer also benefits from an early look at how the process integrates with preceding and subsequent processes and how large is the process variation latitude. In both cases real-time visualization of equipment and process operation offers substantial value. Manufacturing education - in school and on-the-job The need to educate people for advanced manufacturing is clear. In universities this means new course and other activities as well as possible changes in overall program and organization. Education is equally important for people already working in manufacturing in industry. If they do not understand the processes and equipment they use - at least at a phenomenological, predictive level - they cannot provide effective manufacturing. Simulation and visualization may be helpful avenues to such goals. The "technology transfer" issue Technology transfer is indeed a major challenge for industrial competitiveness within the research-development-manufacturing community in industry. Its role is more subject to debate as regards the university and government laboratory research communities. Industry looks to universities, for example, first for people transfer (i.e., the education of scientists and engineers with higher-order skills to impact manufacturing), second for knowledge transfer to enhance the industrial research basis/portfolio, and only third for transfer of specific technology elements The choice of research and educational directions in the university is therefore much more critical in manufacturing competitiveness than is technology transfer between university and industry. Bridging the science/manufacturing gap To exploit the U.S.'s outstanding research base for manufacturing competitiveness requires two-way interaction between science and manufacturing environments. First, the science/university community needs to appreciate what manufacturing is, where scientific research can supply crucial knowledge relevant to manufacturing (especially manufacturability), and with what vehicles close cooperation can be established with industry. Second, the industry/manufacturing community needs to achieve a degree of familiarity with research trends and opportunities sufficient that meaningful discourse can occur in order to encourage those research directions most likely to enhance manufacturing. Semi-tutorial exercises at conferences are one way to build bridges across this gap. Topical Conference exercise on Manufacturing Science We are currently planning a Topical Conference for the 11/93 American Vacuum Society meeting on the subject of manufacturing science, an attempt to assist the redirection of surface science, thin film materials, and vacuum technology research to key issues for manufacturability (primarily in Si microelectronics). The primary subject areas are: defect identification/control and surface cleaning; 3-D and selective processing; process integration and advanced equipment; process/material characterization and real-time control; and selected other aspects of efficient manufacturing (e.g., 6-sigma). In revealing areas of opportunity relevant to current expertise in the AVS community, we hope also to reinforce some central, transcendent themes, such as: (i) turn-around-time for learning is critical from research through manufacturing; (ii) the higher-level viewpoint must be central, for it defines which problems are most important (e.g., show-stoppers) and what elements must be integrated and optimized together; (iii) equipment is at least as important as the process it is intended to accomplish; (iv) technology modeling offers substantial potential, particularly if the more subtle areas of its application are identified; (v) detect control means both isol! ated and parametric defects, and both particles and reactive atomic impurities; (vi) chemical/physical/thermal processes must be understood in 3-dimensional context of microelectronic structures and macro-scale geometry of process equipment; and (vii) rate-limiting steps in chemical reaction paths, as well as competing reaction pathways/kinetics, are crucial to high quality materials and structures with low defect levels. University infrastructure issues The interdisciplinary character of manufacturing science/research tests the flexibility of university structures and programs. Awareness of other technical areas outside the student's specialty, as well as some degree of familiarity with fields like business, requires a different distribution of student time and focus, consistent with engineering education reform efforts currently under way which would create a new breed of engineer. Implementation of interdisciplinary education and research is also a challenge for the university, where organizational structures and course offerings typically follow historically-derived disciplinary/departmental lines. While major restructuring of university organization and procedures is not necessary, some central issues will need to be addressed, e. g. departmental evaluation and promotion decision-making processes for faculty members strongly involved in interdisciplinary research across department and even college boundaries Industry infrastructure issues Current economic stresses and increasing competitiveness of the marketplace are forcing a reduction in industrial support for exploratory research of the kind which could provide strategic leverage to manufacturing competitiveness. This has at least two serious consequences. First, national research investment will decrease due to industry cutbacks, both research projects in industry and contract support from industry to universities. Second, and perhaps more profound, if fewer scientists/researchers (numbers and percentages) are present in industry, communication between industry and university (or government labs) will become harder - precisely the wrong effect given the challenge of bridging the science/manufacturing gap. Approaches are needed to offset these negative factors, e g funding of real industry-university research collaborations. Finally, close interactions between industry (especially manufacturing) and the external R&D community are necessary to attract young people to manufacturing as an exciting enterprise (as well as to encourage manufacturing-relevant research), to enhance the levels of recognition and reward offered in manufacturing, and thereby to elevate the level of education and technical talent in manufacturing Research infrastructure issues and the role of NSF Research relevant to manufacturing demands in many cases sophisticated technical capability. While it is not realistic for university research to measure defect-limited yields, it is crucial to have in place what's needed to elucidate mechanisms which control yield and performance. Consider the earlier example of metal CVD. A crucial manufacturability issue is conformality of the grown film, i.e. how well it covers sidewalls of 3-D structures on a submicron scale. Researchers attracted to this critical manufacturability issue need to have substrates containing deep trench structures, and they must also be able to carry out other processes in order to fabricate meaningful electrical/device test structures to correlate process/tool performance to product quality. It is unreasonable to expect many university or government laboratories to build up sophisticated process lines for trench and then device fabrication. However, central agencies like NSF could play an extremely valuable role by assuring a source for such fabrication - i.e., a single laboratory whose missio! n included building standard test and device structures on wafers which were then available to the entire research community. Another role might be to encourage standardization of wafer sizes so that different research groups could exchange wafers readily for comparison tests or for combinations of processes available only in different laboratories. The effectiveness of the U S. research community toward manufacturing competitiveness will depend on how well we organize, centralize, and share common infrastructure needs. K-12 education A broader infrastructure issue of paramount importance concerns the supply of talent available for engineering and manufacturing research, which is largely determined by education at the precollege level. Developing programs and curricula which encourage interest and accomplishment in science and technology-related K-12 arenas should be a high national priority in order to later yield creative and talented scientists and engineers, as well as a much more technology-literate citizenry. Because it is so relevant to young people's lives, technology may be the preferred approach to engendering interest in science. However, major challenges lie ahead to accomplish such fundamental change in the K-12 scene, particularly in establishing interactions with the K-12 education community which provide new perspectives, programs, and materials to that community. NOTES FOR NSF WORKSHOP Thomas Sciance Conoco (a DuPont subsidiary) Needs exist in "Technical Core Competencies": o Chemical Science and Catalysis o Polymer Synthesis and Science o Coatings o Fiber Technology o Imaging Applications o Plant Science o Fine Particle Technology o Petroleum Technology o Manufacturing Facilities and Systems Technologies o Computational Science (Emerging) o Biotechnology (Emerging) For each Core Competency, a "White Paper" will be developed in 1993 by the technology stewards, covering the following items: o Overall strategy for the competency area o State of the competency area and supporting technologies o Gaps, needs and issues o Comparison to competition o CS&E (central organization) support level o Recommended plans o Major accomplishments from the past year o Milestones for the coming year One version of DuPont's list of "Supporting Technologies" for the core competence of "Manufacturing Facilities and Systems": o Reaction Engineering o Process Measurement o Process Control o Modeling Technologies o Environmental Engineering o Materials Engineering o Manufacturing Systems Engineering Another version of the list: o Particle Technology o Process Synthesis o Process Modeling o Process Sensors and Analyzers o Reaction Engineering o Polymerization o Materials of Construction o Process Control o Analytical Problems: Many technologies appear in polymers, fibers and chemicals "core competency" areas that overlap with those in manufacturing. This makes coordination of the effort and assignment of responsibility for "stewardship" difficult and confusing. The definition of a "core competency" is the ability of an organization to combine a group of technologies to bring products to the marketplace. Since manufacturing systems is a cluster of technologies that does not in itself bring products to the marketplace, it doesn't fit the definition very well. On the other had, it clearly calls for special emphasis. There is a meeting in Wilmington this week (December 2) to discuss this issue. SOME THOUGHTS ON STUFF AND RESEARCH NEEDS IN MANUFACTURING Jeff Siirola Eastman Chemical Company Manufacturing is often associated with "things" and also with the means for creating and producing "things". The processing industries, on the other hand, are concerned with "stuff" from which "things" and fashioned, and also with the means for creating and producing "stuff". Things and the means for producing things are in many ways fundamentally different from stuff and the means for producing stuff. With the exceptions of food, water, and fuel, the end-customer deals far more with things than stuff. Performance (availability, fitness-for-use, safety, cost, etc.) of things is more of a concern to the end-consumer than generally is the performance of stuff. And the broad end-consumer free marketplace serves as arena in which the performance of things evolves at a rapid pace. The marketplace for stuff, although fiercely competitive, is nevertheless very much smaller than the marketplace for things, and I think the process of optimization by evolution much slower. The manufacture of things is generally labor (or its mechanized equivalent) intensive; the production of stuff is generally raw material and energy intensive. Things can be designed to be safe or their hazards are obvious and well known; stuff is often unobviously hazardous. Most things are fashioned and assembled with little impact on the environment; stuff on the other hand is generally made from other stuff with inevitably some stuff left over which has to go somewhere. Finding out all about natural stuff, and thinking up all kinds of ways to react stuff with stuff to make new (sometimes even useful) stuff has long been supported by academia, industry, and government. And so too has research into why stuff behaves the way it does, in all kinds of environments, including those designed to have stuff do its thing at larger and larger scales. Most of the current Division of Chemical and Thermal Systems programs relate directly to supporting understanding why stuff does its thing - kinetics, catalysis, reaction engineering, active surfaces, interfacial transport, separation processes, fluid flow, and heat transfer are all in this category. Possibly only process design and control is principally concerned producing stuff rather than the fundamental behavior of stuff. R & D in manufacturing has traditionally been the purview of the industrial sector. Recent suggestions that government sponsorship of such "applied" research could also be in the national interest has been met with howls of protest, mostly from those currently being supported to study the purer fundamental nature of things and stuff nervous about sharing an always too small pie. I do not believe that industry alone benefits from research and advances in better means of making things and producing stuff. Clearly the end-consumers of that which was made, and all of society, also share in such benefits. This should be obvious from unbiased observation of other nations which have experimented with broadly supported manufacturing research. Since many of the successful examples have involved applied research into the means of manufacture of things, it could be properly asked, and it is the purpose of the workshop to answer, whether there is any need or reasonable expectation of payoff for similar government supported applied research into the means of production of stuff. I would argue in the affirmative. There will always be contention among the competing forces in any production scheme of performance, schedule, and cost (good, fast, and cheap ... pick any two (any one?)). I believe that one can not resolve this optimization from just fundamental understanding alone. Overall systems approaches are needed. Chemical engineers have prided themselves on their systems view, but we still have a long way to go. And perhaps because of the less appreciated, but possibly greater social impact of producing stuff (because of raw materials and energy consumption and environmental and health impacts), government support of process research is at least as appropriate as its support of other manufacturing, not to mention that the performance of stuff-making processes has itself a direct impact on thing-manufacturing competitiveness. I have been personally involved in process research for some time. From these experiences, I am convinced of the tremendous largely untapped potential to decrease energy consumption (for example through design and control of integrated separation systems with heat recovery and the use of non-traditional separation techniques), to decrease environmental impact (again through process design alternatives to eliminate solvents and mass separation agents, integrated reaction and separation, greater recovery of coproducts, more sophisticated scheduling and control systems, etc.), and to reduce process development cycle time (through automated, statistically designed piloting, advanced computer synthesis, modeling and simulation techniques, more reliable scale-up, etc.). These are all large-scale systems integration issues, quite separate from fundamental engineering understanding, and not particularly well understood themselves. Individual successes, some derived from the limited process theory research done in academia and others from industrial efforts, indicate to me great potential to make a real impact on the effectiveness of producing stuff. Successes thus far have been mostly ad hoc. We have a long way to go to better understand the best ways to go about stuff making. This initiative could be a good place to start. MANUFACTURING WORKSHOP DIVISION OF CHEMICAL AND THERMAL SYSTEMS NATIONAL SCIENCE FOUNDATION Irving G. Snyder, Jr. The Dow Chemical Company If the Division of Chemical and Thermal Systems of the NSF wishes to enhance the competitive standing of the manufacturing plants in the continuous process industries in the U.S. I would suggest that the first step of this process would be to establish some desirable attributes or improvements in these manufacturing plants. A few suggested generic improvements are: 1. better control, 2. more highly automated, 3. higher quality product, 4. more energy efficient, 5. fewer environmental insults, and 6. more reliable operations. If this list of desired improvements could be agreed upon. then the next step might be to determine which research and development programs should be encouraged and supported that would help realize these improvements. A suggested way of presenting and prioritizing this information follows on the next page. THE ROLE OF THE CHEMICAL AND THERMAL SYSTEMS DIVISION IN IMPROVING CHEMICALS MANUFACTURING Randall E. Varilek Air Products & Chemical Inc. All Chemical Manufacturers are facing increased concerns from Federal, State, and Local Governments; from environmental groups; and from citizens that live within the proximity of chemical plants. These concerns are threatening the growth and the very existence of the Chemical Industry in this country. If the Chemical Industry is going to survive and thrive, it needs to address these concerns by minimizing its impact upon the environment and ensuring that its processes operate incident free. Both of these areas could see significant benefits from the research supported by the Chemical and Thermal Systems Division. The environmental impact from chemical processes can be minimized by improving the processes with catalysts or new unit operations and with improvements to the end-of-pipe treatment of wastes. The operations safety can be improved through advances in on-line instrumentation and improved information handling techniques to reduce human error while handling th! e onslaught of information generated during nonroutine or emergency situations. {PAGE|3}